x86/sgx: Periodically yield in EPC sanitization to unblock rcu_tasks GP
PATCH 1/1 x86/sgx: Periodically yield in EPC sanitization to unblock rcu_tasks GP
During early boot, ksgxd (Intel Software Guard Extensions Kernel Thread) iterates over all post-kexec dirty EPC pages in a tight loop calling cond_resched() after each page. But, on isolated CPUs
(a common configuration in cloud VMs), cond_resched() never triggers a real context switch because TIF_NEED_RESCHED is not set when no competing runnable task exists on that CPU.
synchronize_rcu_tasks(), invoked by BPF LSM during initialization, must wait for every task that was running at the start of the grace period to pass through a quiescent state (a voluntary sleep or preemption point). If ksgxd never leaves the CPU, the rcu_tasks grace period stalls, causing boot delays exceeding 60 seconds on machines with large EPC regions.
Fix this by introducing SGX_SANITIZE_RESCHED_INTERVAL (32768) and forcing ksgxd to sleep for one jiffy every that many pages, guaranteeing that an rcu_tasks quiescent state is reached in bounded time regardless of CPU isolation. Keep cond_resched() for all other iterations.
Without this patch, instead, virtual machines (VMs) experience a long OS boot times:
[ 4.110549] systemd[1]: Detected architecture x86-64. [ 4.115279] systemd[1]: Hostname set to <i2bp1g0g0m0i8406er0g1zX2>. [ 4.115554] systemd[1]: Installed transient /etc/machine-id file. [ 14.262158] rcu_tasks_wait_gp: rcu_tasks grace period number 1 (since boot) is 10087 jiffies old. [ 14.374158] rcu_tasks_wait_gp: rcu_tasks grace period number 1 (since boot) is 40199 jiffies old. [ 134.806157] rcu_tasks_wait_gp: rcu_tasks grace period number 1 (since boot) is 130631 jiffies old. [ 248.086158] INFO: task systemd:1 blocked for more than 122 seconds. [ 248.086491] Not tainted 6.8.0-90-generic #91-Ubuntu [ 248.086739] 'echo 0 > /proc/sys/kernel/hung_task_timeout_secs' disables this message. [ 248.086993] task:systemd state:D stack:0 pid:1 tpid:1 ppid:0 flags:0x00000002 [ 248.087274] Call Trace: [ 248.087434] <TASK> [ 248.087557] __schedule+0x27c/0x6b0 [ 248.087770] schedule+0x33/0x110 [ 248.087939] schedule_timeout+0x157/0x170 [ 248.088120] wait_for_completion+0x88/0x150 [ 248.088304] __wait_rcu_gp+0x17e/0x190 [ 248.088481] synchronize_rcu_tasks_generic+0x64/0x60 [ 248.088672] ? __pfx_call_rcu_tasks+0x10/0x10 [ 248.088858] ? __pfx_wakeme_after_rcu+0x10/0x10 [ 248.089047] synchronize_rcu_tasks+0x15/0x20 [ 248.089260] register_ftrace_direct+0x31f/0x350 [ 248.089445] ? __pfx_bpf_lsm_file_open+0x10/0x10 [ 248.089629] bpf_trampoline_update+0x469/0x650 [ 248.089814] ? 0xffffffffffffffff [ 248.089988] ? 0xffffffffffffffff [ 248.090153] __bpf_trampoline_link_prog+0x10d/0x330 [ 248.090339] bpf_trampoline_link_prog+0x33/0x60 [ 248.090518] bpf_tracing_prog_attach+0x3c5/0x5f0 [ 248.090699] link_create+0x1a5/0x280 [ 248.090886] ? security_bpf+0x3c/0x70 [ 248.091101] __sys_bpf+0x4ae/0x10 [ 248.091312] __x64_sys_bpf+0x1a/0x30 [ 248.091477] x64_sys_call+0x199/0x250 [ 248.091647] do_syscall_64+0x7f/0x180 [ 248.091818] ? arch_exit_to_user_mode_prepare.isa.0+0x1a/0x60 [ 248.092022] ? irqentry_exit_to_user_mode+0x38/0x1e0 [ 248.092246] ? irqentry_exit+0x43/0x50 [ 248.092401] entry_SYSCALL_64_after_hwframe+0x78/0x80 [ 248.092590] RIP: 0033:0x7b53e592728d [ 248.092756] RSP: 002b:00007ffdaa9d696 EFLAGS: 00000246 ORIG_RAX: 0000000000000141 [ 248.092856] RAX: ffffffffffffffda RBX: 00007ffdaa9d696 RCX: 00007b53e592728d [ 248.092956] RDX: 0000000000000000 RSI: 00007ffdaa9d696 RDI: 0000000000000001 [ 248.093056] RBP: 00007ffdaa9d696 R08: 00007b53e5a03a8 R09: 00007ffdaa9d696 [ 248.093156] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000 [ 248.093256] R13: 0000000000000000 R14: 00005d81ed2cfd0 R15: 00005d81ed2b7ec0 [ 248.093406] </TASK>
--- arch/x86/kernel/cpu/sgx/main.c | 22 ++++++++++++++++++++-- 1 file changed, 20 insertions(+), 2 deletions(-)
arch/x86/kernel/cpu/sgx/main.c +20 -2
@@ -52,6 +52,13 @@ static struct sgx_numa_node *sgx_numa_nodes;static LIST_HEAD (sgx_dirty_page_list); +/* +* Force a voluntary context switch every SGX_SANITIZE_RESCHED_INTERVAL +* iterations to let synchronize_rcu_tasks() (e.g. called by BPF LSM at +* init) complete its grace period. +*/ +#define SGX_SANITIZE_RESCHED_INTERVAL (1 << 15) +/* * Reset post-kexec EPC pages to the uninitialized state. The pages are removed * from the input list, and made available for the page allocator. SECS pages
@@ -63,6 +70,7 @@ static LIST_HEAD(sgx_dirty_page_list);static unsigned long __sgx_sanitize_pages (struct list_head *dirty_page_list) {unsigned long left_dirty =0 ; +unsigned long count =0 ;struct sgx_epc_page *page;LIST_HEAD (dirty);int ret;
@@ -72,6 +80,18 @@ static unsigned long __sgx_sanitize_pages(struct list_head *dirty_page_list)if (kthread_should_stop ())return 0 ; +/* +* On isolated CPUs cond_resched() does not trigger a real +* context switch when no competing runnable task exists. +* Periodically force ksgxd to sleep so that synchronize_rcu_tasks() +* (e.g. BPF LSM) can complete the grace period in bounded time. +* Keep cond_resched() between forced sleeps for higher-priority tasks. +*/ +if (!(++count & (SGX_SANITIZE_RESCHED_INTERVAL -1 ))) +schedule_timeout_interruptible (1 ); +else +cond_resched (); + page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);/*
@@ -105,8 +125,6 @@ static unsigned long __sgx_sanitize_pages(struct list_head *dirty_page_list)list_move_tail (&page->list, &dirty); left_dirty++; } - -cond_resched (); }list_splice (&dirty, dirty_page_list);
Signature · 1 line
2.43.0
Jarkko Sakkinen
maintainerQuoted 83 lines · Jun Miao
During early boot, ksgxd (Intel Software Guard Extensions Kernel Thread) iterates over all post-kexec dirty EPC pages in a tight loop calling
cond_resched()after each page. But, on isolated CPUs
(a common configuration in cloud VMs),cond_resched()never triggers a real context switch becauseTIF_NEED_RESCHEDis not set when no competing runnable task exists on that CPU.
synchronize_rcu_tasks(), invoked by BPF LSM during initialization, must wait for every task that was running at the start of the grace period to pass through a quiescent state (a voluntary sleep or preemption point). If ksgxd never leaves the CPU, thercu_tasksgrace period stalls, causing boot delays exceeding 60 seconds on machines with large EPC regions.Fix this by introducing
SGX_SANITIZE_RESCHED_INTERVAL(32768) and forcing ksgxd to sleep for one jiffy every that many pages, guaranteeing that anrcu_tasksquiescent state is reached in bounded time regardless of CPU isolation. Keepcond_resched()for all other iterations.Without this patch, instead, virtual machines (VMs) experience a long OS boot times:
[ 4.110549] systemd[1]: Detected architecture x86-64. [ 4.115279] systemd[1]: Hostname set to <i2bp1g0g0m0i8406er0g1zX2>. [ 4.115554] systemd[1]: Installed transient /etc/machine-id file. [ 14.262158] rcu_tasks_wait_gp: rcu_tasks grace period number 1 (since boot) is 10087 jiffies old. [ 14.374158] rcu_tasks_wait_gp: rcu_tasks grace period number 1 (since boot) is 40199 jiffies old. [ 134.806157] rcu_tasks_wait_gp: rcu_tasks grace period number 1 (since boot) is 130631 jiffies old. [ 248.086158] INFO: task systemd:1 blocked for more than 122 seconds. [ 248.086491] Not tainted 6.8.0-90-generic #91-Ubuntu [ 248.086739] 'echo 0 > /proc/sys/kernel/hung_task_timeout_secs' disables this message. [ 248.086993] task:systemd state:D stack:0 pid:1 tpid:1 ppid:0 flags:0x00000002 [ 248.087274] Call Trace: [ 248.087434] <TASK> [ 248.087557] __schedule+0x27c/0x6b0 [ 248.087770] schedule+0x33/0x110 [ 248.087939] schedule_timeout+0x157/0x170 [ 248.088120] wait_for_completion+0x88/0x150 [ 248.088304] __wait_rcu_gp+0x17e/0x190 [ 248.088481] synchronize_rcu_tasks_generic+0x64/0x60 [ 248.088672] ? __pfx_call_rcu_tasks+0x10/0x10 [ 248.088858] ? __pfx_wakeme_after_rcu+0x10/0x10 [ 248.089047] synchronize_rcu_tasks+0x15/0x20 [ 248.089260] register_ftrace_direct+0x31f/0x350 [ 248.089445] ? __pfx_bpf_lsm_file_open+0x10/0x10 [ 248.089629] bpf_trampoline_update+0x469/0x650 [ 248.089814] ? 0xffffffffffffffff [ 248.089988] ? 0xffffffffffffffff [ 248.090153] __bpf_trampoline_link_prog+0x10d/0x330 [ 248.090339] bpf_trampoline_link_prog+0x33/0x60 [ 248.090518] bpf_tracing_prog_attach+0x3c5/0x5f0 [ 248.090699] link_create+0x1a5/0x280 [ 248.090886] ? security_bpf+0x3c/0x70 [ 248.091101] __sys_bpf+0x4ae/0x10 [ 248.091312] __x64_sys_bpf+0x1a/0x30 [ 248.091477] x64_sys_call+0x199/0x250 [ 248.091647] do_syscall_64+0x7f/0x180 [ 248.091818] ? arch_exit_to_user_mode_prepare.isa.0+0x1a/0x60 [ 248.092022] ? irqentry_exit_to_user_mode+0x38/0x1e0 [ 248.092246] ? irqentry_exit+0x43/0x50 [ 248.092401] entry_SYSCALL_64_after_hwframe+0x78/0x80 [ 248.092590] RIP: 0033:0x7b53e592728d [ 248.092756] RSP: 002b:00007ffdaa9d696 EFLAGS: 00000246 ORIG_RAX: 0000000000000141 [ 248.092856] RAX: ffffffffffffffda RBX: 00007ffdaa9d696 RCX: 00007b53e592728d [ 248.092956] RDX: 0000000000000000 RSI: 00007ffdaa9d696 RDI: 0000000000000001 [ 248.093056] RBP: 00007ffdaa9d696 R08: 00007b53e5a03a8 R09: 00007ffdaa9d696 [ 248.093156] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000 [ 248.093256] R13: 0000000000000000 R14: 00005d81ed2cfd0 R15: 00005d81ed2b7ec0 [ 248.093406] </TASK>Reported-by challvy <[email protected]>Fixes e7e0545299d8 ("x86/sgx: Initialize metadata for Enclave Page Cache (EPC) sections")Co-developed-by Fan Du <[email protected]>Signed-off-by Fan Du <[email protected]>Signed-off-by Jun Miao <[email protected]>--- arch/x86/kernel/cpu/sgx/main.c | 22 ++++++++++++++++++++-- 1 file changed, 20 insertions(+), 2 deletions(-)arch/x86/kernel/cpu/sgx/main.c +0
-0
arch/x86/kernel/cpu/sgx/main.c static struct sgx_numa_node *sgx_numa_nodes; · line 60
Quoted 9 lines
arch/x86/kernel/cpu/sgx/main.c +6
-0@@ -52,6 +52,13 @@ static struct sgx_numa_node *sgx_numa_nodes;static LIST_HEAD (sgx_dirty_page_list); +/* +* Force a voluntary context switch every SGX_SANITIZE_RESCHED_INTERVAL +* iterations to let synchronize_rcu_tasks() (e.g. called by BPF LSM at +* init) complete its grace period. +*/ +#define SGX_SANITIZE_RESCHED_INTERVAL (1 << 15)
nit: maybe SGX_SANITIZE_PAGES_RECHED_INTERVAL
arch/x86/kernel/cpu/sgx/main.c static LIST_HEAD(sgx_dirty_page_list); · line 76
Quoted 12 lines
arch/x86/kernel/cpu/sgx/main.c +2
-0+/* * Reset post-kexec EPC pages to the uninitialized state. The pages are removed * from the input list, and made available for the page allocator. SECS pages @@ -63,6 +70,7 @@ static LIST_HEAD(sgx_dirty_page_list);static unsigned long __sgx_sanitize_pages (struct list_head *dirty_page_list) {unsigned long left_dirty =0 ; +unsigned long count =0 ;struct sgx_epc_page *page;LIST_HEAD (dirty);int ret;
nit: reverse xmas tree order
Quoted 32 lines
arch/x86/kernel/cpu/sgx/main.c +12
-2@@ -72,6 +80,18 @@ static unsigned long __sgx_sanitize_pages(struct list_head *dirty_page_list)if (kthread_should_stop ())return 0 ; +/* +* On isolated CPUs cond_resched() does not trigger a real +* context switch when no competing runnable task exists. +* Periodically force ksgxd to sleep so that synchronize_rcu_tasks() +* (e.g. BPF LSM) can complete the grace period in bounded time. +* Keep cond_resched() between forced sleeps for higher-priority tasks. +*/ +if (!(++count & (SGX_SANITIZE_RESCHED_INTERVAL -1 ))) +schedule_timeout_interruptible (1 ); +else +cond_resched (); + page = list_first_entry(dirty_page_list, struct sgx_epc_page, list);/* @@ -105,8 +125,6 @@ static unsigned long __sgx_sanitize_pages(struct list_head *dirty_page_list)list_move_tail (&page->list, &dirty); left_dirty++; } - -cond_resched (); }list_splice (&dirty, dirty_page_list);Signature · 1 line
2.43.0
BR, Jarkko
Quoted 13 lines · Jun Miao
During early boot, ksgxd (Intel Software Guard Extensions Kernel Thread) iterates over all post-kexec dirty EPC pages in a tight loop calling
cond_resched()after each page. But, on isolated CPUs
(a common configuration in cloud VMs),cond_resched()never triggers a real context switch becauseTIF_NEED_RESCHEDis not set when no competing runnable task exists on that CPU.
synchronize_rcu_tasks(), invoked by BPF LSM during initialization, must wait for every task that was running at the start of the grace period to pass through a quiescent state (a voluntary sleep or preemption point). If ksgxd never leaves the CPU, thercu_tasksgrace period stalls, causing boot delays exceeding 60 seconds on machines with large EPC regions.
Quoted 6 lines
Fix this by introducing
SGX_SANITIZE_RESCHED_INTERVAL(32768) and forcing ksgxd to sleep for one jiffy every that many pages, guaranteeing that anrcu_tasksquiescent state is reached in bounded time regardless of CPU isolation. Keepcond_resched()for all other iterations.Without this patch, instead, virtual machines (VMs) experience a long OS boot times:
This seems a common problem at the scheduler and RCU layer, but not specific to SGX. I originally thought cond_resched() should be able to handle this, but after reading some history it is not the case.
IIUC, cond_resched_tasks_rcu_qs() was specifically introduced to to address this issue. See bde6c3aa9930 ("rcu: Provide cond_resched_rcu_qs() to force quiescent states in long loops").
Could you try this solution?
Quoted 9 lines
This seems a common problem at the scheduler and RCU layer, but not specific to SGX. I originally thought
cond_resched()should be able to handle this, but after reading some history it is not the case.IIUC,
cond_resched_tasks_rcu_qs()was specifically introduced to to address this issue. Seebde6c3aa9930("rcu: Providecond_resched_rcu_qs()to force quiescent states in long loops").Could you try this solution?
Hi, Kai
The result is effective, and I will adopt this more common method you suggested by, submit the v2 version.
Thanks Challvy Tee for validate it as follow:
We have tested the suggested fix (replacing cond_resched() with cond_resched_tasks_rcu_qs()) on our production environment and can confirm it resolves the issue.
Test environment:
- Instance type: ecs.r9it.16xlarge (SGX-enabled cloud VM with isolated CPUs) on ECS
- Kernel: 6.6.102-cbp.git.772159d5f.alnx4.x86_64 (PREEMPT_DYNAMIC)
Results:
- Before fix: boot time ~50s (with rcu_tasks grace period stall)
- After fix: boot time ~10.7s (systemd-analyze: 724ms kernel + 1.575s initrd + 8.481s userspace = 10.782s)
```
# uname -a
Linux iZbp1cp63g07kzbp93uzrhZ 6.6.102-cbp.git.772159d5f.alnx4.x86_64 #1 SMP PREEMPT_DYNAMIC Mon Jun 22 03:10:04 UTC 2026 x86_64 x86_64 x86_64 GNU/Linux
# free -h
total used free shared buff/cache available
Mem: 243Gi 4.5Gi 239Gi 2.3Mi 481Mi 239Gi
Swap: 0B 0B 0B
# systemd-analyze
Startup finished in 724ms (kernel) + 1.575s (initrd) + 8.481s (userspace) = 10.782s
multi-user.target reached after 3.939s in userspace.
# systemd-analyze blame | head
5.416s kdump.service
2.743s sys-devices-pci0000:00-0000:00:05.0-virtio1-net-eth0.device
2.743s sys-subsystem-net-devices-eth0.device
2.534s sys-devices-pci0000:00-0000:00:03.0-virtio0-virtio\x2dports-vport0p1.device
2.534s dev-virtio\x2dports-org.qemu.guest_agent.0.device
2.534s dev-vport0p1.device
2.512s sys-module-configfs.device
2.510s dev-ttyS0.device
2.510s sys-devices-pnp0-00:00-tty-ttyS0.device
2.509s sys-devices-platform-serial8250-tty-ttyS1.device
# grep -o 'sgx[^ ]*' /proc/cpuinfo | sort -u
sgx
sgx_lc# dmesg | grep -i sgx [ 0.355547] sgx: EPC section 0x4040000000-0x7e3fffffff
# ls -la /dev/sgx* crw-rw---- 1 root sgx 10, 125 Jun 22 11:36 /dev/sgx_enclave crw------- 1 root root 10, 126 Jun 22 11:36 /dev/sgx_provision
Warm regards
Jun Miao